In this paper, an inter-turn short-circuit (ITSC) fault detection method is presented based on the wavelet packet decomposition and locus of three-phase currents in 3-D space for squirrel cage induction motors (SCIMs). The proposed method is able to accurately identify the fault severity and location under any load condition and any type of motor power supply. A comprehensive dynamic model is also proposed for SCIMs having ITSC fault in one phase or a combination of multiple phases. The proposed signal-based fault detection method is separately evaluated based on the data extracted from the proposed model, from the finite element method (FEM) and from the experimental test-rig which consists of two 1.5 kW induction motors, a 3 kW self-excitation generator, three LA55-P/SP1 Hall effect current sensors, a TMS320F28379D digital signal processor (DSP) and an electric drive system. The efficacy of the proposed fault detection method is evident from the results.